IMPLEMENTASI KRIPTANALIS BERBASIS NEURAL NETWORK FEEDFORWARD BACKPROPAGATION MULTI LAPIS TERHADAP KRIPTOGRAFI S-DES

Abstract

Cryptography is used to ensure the confidential information between the sender and the recipient. The sender encrypts the confidential information before it is transmitted to the recipient. After the encrypted confidential information is received by the recipient, the recipient then performs decryption on the encrypted confidential information. In practice, there are third parties who attempt to use various methods to find out the contents of the confidential information. In this study, the researchers perform cryptanalysis on the cryptographic system. The researchers use the S-DES cryptographic method and the neural network multilayer backpropagation cryptanalysis method. Both the training and testing processes use all possible combinations of plaintexts, keys, and ciphertexts in the S-DES cryptographic system. The accuracy of the trained model is evaluated by calculating the match percentage between ciphertext that produced by the model and ciphertext that produced by S-DES. The results show the neural network–based cryptanalysis that used in this research is not sufficiently accurate in performing cryptanalysis on the S-DES cryptographic system. This can be seen from the accuracy test results, which are below 1%.